Week 35 · Unit 19: Interpreting and discussing results · 4 lessons
4
Comparing distributions and drawing conclusions
Coursebook 9: Unit 19, pp. 184–189
GoalCompare two distributions from graphs and statistics, say which claims the data supports, and write a conclusion that answers the original question.
New words
distribution · the way data is spread across its possible valuessupports · gives good evidence for a statement; evidence can also cast doubt on a statementconclusion · a final statement that answers the question, backed by evidence
Explanation
To compare two distributions, look at the shape of the graphs. Where is the peak? Is one set higher overall? Is one set wider, which means more spread out? Support each comment with numbers, such as a mean, a median, a range, or how many values are above a given value. If the groups have different sizes, compare percentages. When you have a claim to test, ask what evidence would support it and what would cast doubt on it. Then write a conclusion that answers the question you started with. Be honest about limits: a small sample, or only two groups, may not show what is true for everyone. Correlation does not prove that one thing causes the other.
Worked examples
The polygons show the ages of 60 runners in a fun run and 60 in a marathon, in classes of width 10 starting at 10 ≤ a < 20 (midpoints 15, 25, 35, …). The fun run peaks at 30 ≤ a < 40 with 18 runners; the marathon peaks at 40 ≤ a < 50 with 20 runners. Estimated means: fun run 34.3 years, marathon 42.7 years. Aged 50 or over: fun run 8 of 60 (13%), marathon 16 of 60 (27%). Conclusion: the marathon runners are older, about 8 years on average, and twice as large a share are 50 or over.
A scatter graph shows the age of 8 phones (years) and their battery life (hours). Claim: older phones have shorter battery lives. The points fall from left to right close to a straight line, so there is strong negative correlation. The line passes through (0, 15.5) and (4.5, 6.1): its gradient is about −2.1, so each extra year of age goes with roughly 2 hours less battery life. The claim is supported. The limits: only 8 phones were tested, and the graph does not show that age alone is the cause, because phone models and use also differ.
Day
Mean
Median
Range
Monday
8.4 min
7 min
14 min
Friday
6.1 min
6 min
4 min
A school looks at its lunch queue times (minutes) for ten Mondays and ten Fridays. Mondays: 12, 5, 9, 3, 14, 6, 8, 17, 4, 6. Fridays: 5, 7, 6, 4, 8, 6, 7, 5, 6, 7. Claim: the queue is shorter and more predictable on Fridays. Evidence: Monday mean 8.4, median 7, range 14; Friday mean 6.1, median 6, range 4. Fridays have the lower mean and median, so they are shorter on average, and the range is much smaller, so they are more predictable. The claim is supported by the data, for these 20 days.
Class activity
Claim detective. Choose a claim you have heard, such as “children today sleep less than their grandparents” or “taller people have bigger feet”. Decide what data you need and collect at least 10 values for each group. Show them in a diagram with one group’s leaves on each side of the stems, or in a scatter graph. Work out the statistics you need. Write a conclusion that says whether the data supports the claim and mentions one limit of your evidence.
Practice
1
The polygons show the ages of 55 people at an animated film (film X) and 55 at a thriller (film Y), in classes of width 10 from 0 ≤ a < 10 (midpoints 5, 15, …, 65). Film X frequencies: 14, 16, 5, 6, 9, 4, 1. Film Y: 1, 2, 10, 18, 14, 8, 2. Write three sentences comparing the two audiences, with numbers.
Film X’s audience is mostly children: 30 of 55 are under 20 and the estimated mean age is about 24. Film Y’s audience is older: its peak is at 30–40 (18 people) and the estimated mean is about 38. Film Y’s ages are more concentrated in the middle; film X has a peak among the youngest and a long tail towards older ages.
2
A scatter graph of 10 students’ hours of sleep and their score in a concentration test shows positive correlation. Which of these statements does the graph support? (a) Students who sleep more tend to score higher. (b) Sleeping more makes every student score higher. (c) A student who sleeps 5 hours will score about 40. Explain.
Only (a) is supported: the points rise from left to right, so there is a general tendency. (b) claims cause for every student, which correlation cannot prove. (c) uses 5 hours, which is outside the data (6 to 9.5 hours), so the prediction is unreliable.
3
The diagram shows the time in seconds that 12 Year 7 and 12 Year 9 students took to read a page. Compare the two years in three sentences using the median and the range.
Year 7: median (35 + 36) ÷ 2 = 35.5 s, range 47 − 28 = 19 s. Year 9: median (27 + 28) ÷ 2 = 27.5 s, range 38 − 21 = 17 s. The Year 9 students read faster, by about 8 seconds on average, and the spreads are similar, so Year 9 is quicker and about as consistent.
4
A survey asks 6 Year 7 and 6 Year 9 students how long they sleep. The Year 9 mean is lower. Mara says: “So all Year 9 students sleep less than Year 7 students.” Explain two reasons why the survey does not show this.
The sample is very small (only 6 in each year), so the result may not be typical of the whole year group. A mean describes a group, not every student; some Year 9 students will sleep longer than some Year 7 students. The way the 12 students were chosen could also be biased.